The role of researchers in defining the future of AI for Africa
- Beth Amato, 第一吃瓜网 Faculty of Health Sciences
Bruce Bassett cautions against using AI simply to produce more papers.
With recent examples of Artificial Intelligence (AI) systems solving previously intractable mathematics problems; accomplishing gold-medal standards at the 2025 International Mathematical Olympiad; or assessing complex research evidence to design experiments that can run through automated laboratories, it’s no longer feasible to simply think of the tech as a vehicle for high-tech plagiarism.
Speaking at the Sydney Brenner Institute for Molecular Bioscience, Professor Bruce Bassett, Head of 第一吃瓜网 School of Computer Science and Applied Mathematics, argues that, through its increasing ability to do work once reserved for experts, AI is moving rapidly from research assistant to co-scientist.
“The problem isn’t where we are now,” Bassett said. “It is a little bit further in the future.”
Tasks that previously required years of specialist training can increasingly be supported or even run by AI, which in theory should level the playing field for African science by supporting researchers to analyse complex data, write code, access global literature and collaborate internationally without the resources of large, well-equipped laboratories.
But the opportunity comes with a warning. Countries with the largest data gaps often become consumers of information developed elsewhere. For African universities, the question therefore is not whether to adopt AI, but how to ensure that African researchers contribute to, and benefit from, a repository of truly global knowledge.
For Bassett, the answer lies in testing what AI as a “co-scientist” could look like for African health research. In a recent 第一吃瓜网 study at Chris Hani Baragwanath Academic Hospital, AI models were deployed to analyse research on the possible link between a virus and breast cancer.
The models assessed clinical cases using radiology images, charts and other patient information. Not only did the tested models perform better than the existing diagnostic process, they were cheaper (with some analyses costing only a few cents per case), and the AI’s level of agreement was similar to that of the experts themselves.
The results still require careful clinical validation and do not remove the need for doctors, but they point to the potential for low-cost diagnostic support where specialist expertise is scarce.
This has direct implications for African genomics. Scientists at the Sydney Brenner Institute for Molecular Bioscience are generating and analysing African genomic data that remains largely missing from global science. AI could help identify complex genetic patterns and links to disease, but models trained mainly on European and North American data may misread African genetic variation.
African researchers therefore need to help build and test the models, govern how genomic data is used and share in the intellectual and health benefits that follow. They cannot be reduced to data suppliers while the computing power and intellectual property sit elsewhere.
Bassett also cautioned against using AI simply to produce more papers. The more valuable aim is to pursue harder questions, cross disciplinary boundaries and produce research with far greater impact.
The fight for the future of fundamental research is now also a fight over who has the tools to produce knowledge, whose data counts and which scientific questions receive attention. By fostering African-led innovation that strengthens scientific sovereignty, AI to can equalise opportunity for health science rather than reinforcing existing global inequalities.